The content of the invention
For existing interference management techniques in performances such as system transfer rate, algorithm complex, control information interaction expenses
The deficiency of aspect, the present invention provide point in the orthogonal frequency-time multiple access network of the optimal low complex degree of a kind of system and speed
Cloth inter-cell downlink interference elimination method.
Realize that the technical solution of the object of the invention is as follows:
Interference elimination method between a kind of distributing cell in orthogonal frequency-time multiple access network, if:Iteration time slot t=1,
2 ..., T (T is maximum iteration), useWithRepresent that base station i adjusts in the transimission power of t iteration time slots and user respectively
Degree strategy, i ∈ N, N={ 1,2 ..., N } represent collection of base stations, and N is total number of base,
Step 1, Initialize installation:Iteration time slot t is made equal to 1, its user's scheduling strategy of each Base station initialization, each base station
Subcarrier is randomly assigned to any one communication user of its service respectively, then, each base station random initializtion is respective
Transmit power.One different adjustment priority λ is set at random to each base stationi, λiBase station i adjustment priority is represented,
Step 2, according to interference indexOrder from big to small, the B base stations to base station i interference effects maximum before taking Interference index for base station j to base station i interference effects, j ∈ N { i }, N { i } represent set
N and the difference set of set { i }, BiFor base station i interference neighbors set, index is disturbedCalculation formula beWherein CiThe communication user set of base station i services is represented, | Ci| represent communication user set CiIn element
Number, n ∈ CiAnd n be base station i service any one communication user sequence number, gj,nFor from base station j to communication user n letter
Road power gain, gi,nFor from base station i to communication user n channel power gain,
Step 3, respectively by the priority of each base station in the interference neighbors set of the priority of each base station and each base station
It is compared, if the priority of base station is higher than the priority of each base station in the interference neighbors set of base station, by the base station
It is classified as coordinating base station k, k ∈ Lt, LtCoordination base station k set is represented, carries out interference coordination processing;Otherwise, without interference coordination
Operation,
Step 4, each base station i ∈ N calculate its rate of information throughput in iteration time slot t using shannon formulaWherein W represents channel width, Represent t-th of iteration time slot
When base station i Signal to Interference plus Noise Ratio, N0For noise power, n ∈ CiThe user serviced by base station i, then calculate each coordination base station k
∈LtValue of utilityAnd Coordinate base station k ∈ LtInterference neighbors base station l ∈ BkInformation transfer speed
Rate and by its report give coordinate base station k, BkFor base station k interference neighbors set,
Step 5, power level examination is surveyed and corresponding value of utility calculates:
Step 5.1, in t≤T/2, each coordination base station k ∈ LtGeneral one examination of random selection such as the probability with 1/M is surveyed
Power level Here Pk,maxFor base station k maximum transmission power, M is power
The number of level, η1=0, ηM=1, and η1<η2<...<ηM, each base station i ∈ N, which are calculated, is based on examination power scale levelMost
Excellent user's scheduling strategyAnd subcarrier is given into rate of information throughput highest service user,Table
Show the user n that base station i ∈ N are serviced based on examination power scale levelThe rate of information throughput, then calculate each coordination base station k
∈LtExamination power scale levelCorresponding value of utilityAnd enter step 6.1,
Step 5.2, in t>In T/2, each coordination base station k ∈ LtAll power levels are surveyed in examination successivelyHere Pk,maxFor base station k maximum transmission power, M is the number of power level, η1
=0, ηM=1, and η1<η2<...<ηM, based on examination power scale levelEach base station i ∈ N perform optimal user and dispatch plan
SlightlyAnd subcarrier is given into the most fast service user of the rate of information throughput,Represent that base station i ∈ N are taken
User's n rates of information throughput of business, then calculate each coordination base station k ∈ LtExamination power scale levelCorresponding value of utilityAnd enter step 6.2,
Step 6, transimission power and the renewal of user's scheduling strategy:
Step 6.1, equally distributed random number θ between random generation one [0,1], ifIt is then next
Timeslot power-levelsIt is updated to try power scale levelOtherwise, ifThen next timeslot power-levels are protected
Hold it is constant i.e.Whereinβ is a coefficient and β>0,To coordinate base station k ∈ Lt
Examination power scale levelCorresponding value of utility, after the completion of renewal, by all coordination base station k ∈ LtPriority reduce by 1 grade, no
Carry out interference coordination operation base station w ∈ N LtPriority keep constant, while new priority is reported to adjacent base station;
Then, each base station i ∈ N calculate the best user's scheduling scheme of next time slot based on the power level after renewalAnd subcarrier is given into rate of information throughput highest service user, whereinRepresent base station i ∈ N
The user n serviced next gap information transmission rate based on the power level after renewal, is transferred to step 7,
Step 6.2, each coordination base station k ∈ LtCompare all examination power scale level { η1Pk,max,η2Pk,max,...,ηMPk,maxCorresponding to value of utility size, choose transimission power of the power level as next time slot corresponding to highest value of utility,
After the completion of renewal, by all coordination base station k ∈ LtPriority reduce by 1 grade, without interference coordination operation base station w ∈ N Lt
Priority keep constant, while new priority is reported to adjacent base station;Then, each base station i ∈ N are calculated based on renewal
The best user's scheduling scheme of next time slot of power level afterwardsAnd subcarrier is given into information transfer speed
Rate highest service user, whereinRepresent the user n that are serviced of base station i ∈ N based under the power level after renewal
The one time slot rate of information throughput, is transferred to step 7,
Step 7, algorithm end condition judges:If algorithm performs reach maximum iteration T, algorithm terminates;Otherwise, t is made
=t+1 simultaneously goes to step 3.
It is of the invention significantly the advantages of be:1) present invention is a kind of distributed embodiment, and each base station is as independent
Individual, autonomous carry out computing and interference coordination Developing Tactics, it is not necessary to which the master controller of a wide area carries out centralized Control
And calculating, this distributed operation cause system architecture flattening, avoid the system crash brought by single point failure, system is more
Steadily and surely;2) local (local) information exchange is only needed, each base station only needs to carry out with adjacent base station during interference coordination
Necessary information exchange, because being connected between adjacent base station by wired optical fiber, the communication between them is very easy;3) divide
The embodiment of cloth shares arithmetic operation does distributed variable-frequencypump and calculating in each base station, each base station, has phase
Than in the lower computation complexity of centralized algorithm;4) in step 6, when iteration time slot t is not more than T/2, using Boltzmann
Probability updating rule (parameter beta is similar to the temperature parameter in simulated annealing), if the value of utility that examination power scale level obtains
More than value of utility corresponding to current power strategy, then next time slot will be used as using larger probability selection examination survey level by coordinating base station
Transimission power, deny with greater probability its power policy being kept constant.But that more preferable, poor plan of whichever strategy
Slightly still there is certain probability to be selected to, this may be such that local optimum is easily jumped out in each base station compared to greedy optimizing algorithm
Interference coordination strategy and the interference coordination strategy for finding global optimum, reach system and speed maximizes;5) in step 6, when repeatedly
When being not more than T/2 for time slot t, the Boltzmann probability updating rules of use make algorithm jump out locally optimal solution and search the overall situation
It is optimal;And when iteration time slot t is more than T/2, using optimal response policy update rule, the selection determined in each iteration is best
Strategy, convergence of algorithm speed can be accelerated.
Embodiment
Interference elimination method between a kind of distributing cell in orthogonal frequency-time multiple access network, it is characterised in that set:Repeatedly
For time slot t=1,2 ..., T (T is maximum iteration), useWithTransmission of the base station i in t iteration time slots is represented respectively
Power and user's scheduling strategy, i ∈ N, N={ 1,2 ..., N } represent collection of base stations, and N is total number of base,
Step 1, Initialize installation:Iteration time slot t is made equal to 1, its user's scheduling strategy of each Base station initialization, each base station
Subcarrier is randomly assigned to any one communication user of its service respectively, then, each base station random initializtion is respective
Transmit power, in addition, in order to determine the precedence of each base station policy update, different to the random setting one in each base station
Adjust priority λi, λiBase station i adjustment priority is represented,
Step 2, according to interference indexOrder from big to small, the B base stations to base station i interference effects maximum before taking Interference index for base station j to base station i interference effects, j ∈ N { i }, N { i } represent set
N and the difference set of set { i }, BiFor base station i interference neighbors set, index is disturbedCalculation formula beWherein CiThe communication user set of base station i services is represented, | Ci| represent communication user set CiIn member
Plain number, n ∈ CiAnd n be base station i service any one communication user sequence number, gj,nFor from base station j to communication user n's
Channel power gain, gi,nFor from base station i to communication user n channel power gain,
Step 3, respectively by the priority of each base station in the interference neighbors set of the priority of each base station and each base station
It is compared, if the priority of base station is higher than the priority of each base station in the interference neighbors set of base station, by the base station
It is classified as coordinating base station k ∈ Lt, LtCoordination base station k set is represented, carries out interference coordination processing;Otherwise, grasped without interference coordination
Make,
Step 4, each base station i ∈ N calculate its rate of information throughput in iteration time slot t using shannon formulaWherein W represents channel width, Represent t-th of iteration time slot
When base station i Signal to Interference plus Noise Ratio, N0For noise power, n ∈ CiThe user serviced by base station i, then calculate each coordination base station k
∈LtValue of utilityAnd Coordinate base station k ∈ LtInterference neighbors base station l ∈ BkInformation transfer speed
Rate and by its report give coordinate base station k, BkFor base station k interference neighbors set, it can be seen that by formula, each effect for coordinating base station
WithThe transmission rate sum of the transmission rate of its own and interference neighbors base station is defined as, therefore each coordination base station is in root
It can both consider its own when carrying out policy update according to its effectiveness, can also consider interference neighbors base station.Thus, the effectiveness design method
Embody the thought cooperated with each other, set BkSize embody the number of cooperation base station, and from step 2 | Bk|=B,
Step 5, power level examination is surveyed and corresponding value of utility calculates:
Step 5.1, in t≤T/2, each coordination base station k ∈ LtGeneral one examination of random selection such as the probability with 1/M is surveyed
Power level Here Pk,maxFor base station k maximum transmission power, M is power
The number of level, η1=0, ηM=1, and η1<η2<...<ηM, each base station i ∈ N, which are calculated, is based on examination power scale levelMost
Excellent user's scheduling strategyAnd subcarrier is given into rate of information throughput highest service user,Table
Show the user n that base station i ∈ N are serviced based on examination power scale levelThe rate of information throughput, then calculate each coordination base station k
∈LtExamination power scale levelCorresponding value of utilityAnd enter step 6.1,
Step 5.2, in t>In T/2, each coordination base station k ∈ LtAll power levels are surveyed in examination successivelyHere Pk,maxFor base station k maximum transmission power, M is the number of power level, η1
=0, ηM=1, and η1<η2<...<ηM, based on examination power scale levelEach base station i ∈ N perform optimal user and dispatch plan
SlightlyAnd subcarrier is given into the most fast service user of the rate of information throughput,Represent that base station i ∈ N are taken
User's n rates of information throughput of business, then calculate each coordination base station k ∈ LtExamination power scale levelCorresponding value of utilityAnd enter step 6.2,
Step 6, transimission power and the renewal of user's scheduling strategy:
Step 6.1, equally distributed random number θ between random generation one [0,1], ifIt is then next
Timeslot power-levelsIt is updated to try power scale levelOtherwise, ifThen next timeslot power-levels are protected
Hold it is constant i.e.Whereinβ is a coefficient and β>0,To coordinate base station k ∈ Lt
Examination power scale levelCorresponding value of utility, after the completion of renewal, by all coordination base station k ∈ LtPriority reduce by 1 grade, no
Carry out interference coordination operation base station w ∈ N LtPriority keep constant, while new priority is reported to adjacent base station;
Then, each base station i ∈ N calculate the best user's scheduling scheme of next time slot based on the power level after renewalAnd subcarrier is given into rate of information throughput highest service user, whereinRepresent base station i ∈ N
The user n serviced next gap information transmission rate based on the power level after renewal, is transferred to step 7,
Step 6.2, each coordination base station k ∈ LtCompare all examination power scale level { η1Pk,max,η2Pk,max,...,ηMPk,maxCorresponding to value of utility size, choose transimission power of the power level as next time slot corresponding to highest value of utility,
After the completion of renewal, by all coordination base station k ∈ LtPriority reduce by 1 grade, without interference coordination operation base station w ∈ N Lt
Priority keep constant, while new priority is reported to adjacent base station;Then, each base station i ∈ N are calculated based on renewal
The best user's scheduling scheme of next time slot of power level afterwardsAnd subcarrier is given into information transfer speed
Rate highest service user, whereinRepresent the user n that are serviced of base station i ∈ N based under the power level after renewal
The one time slot rate of information throughput, is transferred to step 7,
Step 7, algorithm end condition judges:If algorithm performs reach maximum iteration T, algorithm terminates;Otherwise, t is made
=t+1 simultaneously goes to step 3, in view of the interference index in step 2It is to evaluate base station j for the flat of the base station i all users serviced
Equal interference effect, interference coordination strategy change is not influenceed the interference relationships analysis in step 2 during by algorithm iteration, therefore
Iterative process is transferred to step 3 by step 7.
The present invention is described in further detail below in conjunction with the accompanying drawings.
Referring to Fig. 1, the inventive method system is provided with the OFDMA system of 49 cells, and each radius of society is set to 1000 meters,
Each center of housing estate is provided with a base station, and random distribution has 8 communication users in each cell.Band system band is divided into 16 sons
Channel, each a width of 200KHz of sub-channel.The maximum transmission power of base station46dBm is set to, and is distributed evenly over its son
On channel.Power level number is set to M=2,4,8.There is identical noise variance σ on each communication link2=-130dBm.
Channel gain model is hi=A/di α, wherein involved parameter is set to:A=0.097, α=3.The greatest iteration of algorithm
Number is set to 400.As Fig. 2, the implementation process of whole example are as follows:
Step 1, Initialize installation:Its subchannel (subcarrier) is randomly assigned to the logical of its service by t=1, each base station
Credit household, it need to ensure only to be taken by a communication user per sub-channels in each cell.Then, each base station of random initializtion
Transmit power on each sub-channels,Hereafter, assisted for all subchannels using following scheme
Adjust the transimission power and user's scheduling strategy between base station.Assuming that each sub-channels are completely orthogonal, therefore on each sub-channels
Interference coordination schemes be separate.In addition, in order to determine the precedence of each base station policy update, to each base station
One different coordination priority λ of setting at randomi∈ [1,2 ..., 49], i ∈ N, N are base station set.
Step 2, interference relationships analysis and interference neighbors set determine:By taking the i ∈ N of any one base station as an example, pass through calculating
Disturb indexCome evaluate other base station j ∈ N { i } to base station i interference effect, wherein N { i } represent set N with gathering
The difference set of { i }.Disturb indexCalculation formula beWherein CiRepresent cell i (base station i service area
Domain) in communication user set, | Ci| represent set CiIn element number, n ∈ CiFor a base station i service user, gj,n
For from base station j to user n channel power gain.WillArrange, obtain from big to smallWhereinRepresent the base station of the interference big to base station i generations b.More strongly disturbing preceding B base station will be produced to base station i and be defined as one
Interference neighbors set,
Step 3, the base station set L that can coordinate simultaneously is determined in a distributed fashiont, ensure mutually strongly disturbing adjacent base station
Interference coordination is asynchronously carried out, i.e.,AndSpecific implementation method is each base station i ∈ N ratios
Compared with the priority of its own and its adjacent base station BiPriority.Have if there is an adjacent base station than it higher preferential
Level, then the base station operates without interference coordination.Otherwise, interference coordination is carried out.
Step 4, each base station i ∈ N calculate its information rate in iteration time slot t using shannon formulaWherein W=200KHz represents channel width,Represent Signal to Interference plus Noise Ratio,
N0For noise power, n ∈ CiThe user serviced by base station i.Each coordination base station k ∈ LtPass through base station l ∈ B adjacent theretok's
Communication obtains its rate of information throughputAnd thus calculate value of utilityWherein BkIt is adjacent for base station k interference
Occupy set.It can be seen that by formula, each effectiveness for coordinating base stationIt is defined as the transmission rate of its own and interference neighbors base station
Transmission rate sum, therefore it is each coordination base station according to its effectiveness carry out policy update when can both consider its own, also can
Consider interference neighbors base station.Thus, the effectiveness design method embodies the thought cooperated with each other, set BkSize embody conjunction
Make the number of base station, and from step 2 | Bk|=B.
Step 5, power level examination is surveyed and corresponding value of utility calculates:
If 1) t≤T/2:Each coordination base station k ∈ LtGeneral one examination power scale level of random selection such as the probability with 1/MHereFor base station k maximum transmission power, M is power level
Number (being set to 2,4,8), η1=0, ηM=1, and η1<η2<...<ηM.Hereafter, based on examination power scale levelIt is each
Base station i ∈ N it is autonomous determine its optimal user's scheduling strategyWhereinRepresent user n be scheduled into
The information rate of row communication.Then, repeat step 4, each coordination base station k ∈ LtCalculate examination power scale levelCorresponding effectiveness
Value
If 2) t>T/2:Each coordination base station k ∈ LtAll power levels are surveyed in examination successively
HereFor base station k maximum transmission power, M is the number (being set to 2,4,8) of power level, η1=0, ηM
=1, and η1<η2<...<ηM.Hereafter, based on examination power scale levelEach base station i ∈ N it is autonomous determine its optimal use
Family scheduling strategyWhereinRepresent the scheduled information rates to be communicated of user n.Then, repeat to walk
Rapid 4, each coordination base station k ∈ LtCalculate examination power scale levelCorresponding value of utility
Step 6, transimission power and the renewal of user's scheduling strategy:
If 1) t>T/2:Each coordination base station k ∈ LtIt is lower for the moment that it is updated according to the following Boltzmann rule of probabilitys
The power level of gap
Whereinβ=t/10 is a positive parameter developed with iteration time,Represent next timeslot power-levelsIt is updated to try power scale levelProbability,Represent
Next timeslot power-levels keep constant probability.Equally distributed random number θ between random generation one [0,1], ifThen next timeslot power-levelsIt is updated to try power scale levelOtherwise, if
Then next timeslot power-levels holding is constant i.e.In addition, remaining non-coordinating base station w ∈ N LtIt is kept to launch work(
Rate is constant, i.e.,Then, its best user is recalculated based on the power level after renewal, each base station i ∈ N
Scheduling schemeWhereinRepresent the scheduled information rates to be communicated of user n.Completed in renewal
Afterwards, all coordination base station k ∈ LtReduce its priority (non-coordinating base station w ∈ N LtPriority keep constant), it is and its is preferential
Level, which is reported, gives its adjacent base station.
If 2) t>T/2:Each coordination base station k ∈ LtCompare all examination power scale level { η1Pk,max,η2Pk,max,...,ηMPk,maxCorresponding to value of utility size, choose transimission power of the power level as next time slot for obtaining highest value of utility, i.e.,
In addition, remaining non-coordinating base station w ∈ N LtKeep its transmission power constant, i.e.,Then, based on more
Power level after new, each base station i ∈ N recalculate its best user's scheduling schemeWhereinRepresent the scheduled information rates to be communicated of user n.After the completion of renewal, all coordination base station k ∈ LtIt is preferential to reduce its
Level (non-coordinating base station w ∈ N LtPriority keep constant), and its priority is reported and gives its adjacent base station.
Step 7, algorithm end condition judges:If algorithm performs reach maximum iteration T (T=400), algorithm terminates;
Otherwise, make t=t+1 and go to step 3.
Fig. 3 show the change curve of system of the invention and rate capability with algorithm iteration number.Depict and work as in figure
The change curve of inventive energy when number of power levels M is respectively 2,4,8 three kinds of schemes.For three kinds of schemes, specifically return
One change power level be respectively { 0,1 }, { 0,1/4,1/2,1 },From figure
As can be seen that the increase of the performance of several schemes all iterationses and improve, and algorithm is restrained after about 150 iteration.I
It is also found that when power level number is more, the performance that the present invention reaches is more excellent.Because more power level numbers
Corresponding to wider array of search solution space, and solution space when solution space during M=8 includes M=4, solution space during M=4 is again
Solution space during comprising M=2.In addition, simulation curve shows power level number by 2 to 4, system and speed have larger carry
Rise, level book further then is brought up into 8 will not bring very big raising.
Fig. 4 show the change curve of system of the invention and speed with cooperation neighbor base stations number B.Depict and work as in figure
The change curve of inventive energy when number of power levels M is respectively 2,4,8 three kinds of schemes.It can be seen that several sides
The performance of case all improves with the increase of cooperation neighbor base stations number.It can be appreciated that cooperation neighbor base stations number is more, interference association
The efficiency of tune is also higher, and the conflict brought each other by independent decision-making is also just smaller.But when cooperation number reaches 8, it is
System and rate capability no longer have a distinct increment.Because general (6-8) base station that only made a circle to week of single base station produce compared with
Big interference, therefore coordinate the neighbouring base station effect of surrounding than more significant.Also it correspond to preferably in addition, power level number is more more
Systematic function.